Radar systems operating at millimeter wavelengths have been in use for some time on vehicles, where they are installed to provide such features as warnings of objects that may not be visible to a driver of the vehicle, adaptive cruise control (ACC), or warnings of the possibility of a collision. Because of their use in vehicles, the radar systems have to comply with a large number of constraints, such as needing to operate in all weather with virtually no on-going maintenance. The systems are very sophisticated, yet need to be priced low enough for a mass market environment. Taken together, the constraints make it difficult to implement a simple, efficient radar transceiver system that is to be usable in a mass market application.
Modem vehicular radars typically require a multi-channel architecture, leading to a complex multi-function antenna and transceiver front end The complexity and stringent requirements on the individual systems typically leads to high cost solutions which limit the applicability of the technology.
Vehicular radar typically operates by transmitting a linearly frequency modulated (J) wave—“a chirp”—and receiving reflected signals from a target. The beat signal between the transmitted waveform and the received signal from a static target has a frequency FR proportional to the range R. If the target is moving relative to the sensor, at a relative velocity, herein termed range rate, {dot over (R)}, then there is an additional contribution FD to the frequency of the beat signal due to the Doppler Effect.fbeat±=FR+FD=±aR+b{dot over (R)}  (1)
where
      a    =                  2        ⁢        B            c        ,      b    =                  2        c            ⁢              f        carrier              ,B is the bandwidth of the chirp, fcarrier is the center frequency of the chirp, and c is the speed of light.
The + and − signs correspond to an upward and downward chirp (positive and negative slope of the LFM) respectively.
When there is a single, well defined target, then both the range and the range rate can be measured by successively transmitting two chirps (triangular waveform), one with upward slope and the second with downward slope. The frequencies of the signals received from the target are measured in each of the chirps, and the range and range rate estimated by adding and subtracting the measured frequencies:
                              R          =                                                    f                beat                +                            -                              f                beat                -                                      a                          ,                              and            ⁢                                                  ⁢                          R              .                                =                                                                      f                  beat                  +                                +                                  f                  beat                  -                                            b                        .                                              (        2        )            
This simple procedure fails when there is more than one target in the field of view. The frequency processing of each chirp results in a number of peaks, each one with its characteristic frequency. The pairing between the peaks in the two spectrums is not unique and in addition to the correct pairing, there exist a multitude of other pairing possibilities that result in non existent targets named ghosts. This situation becomes increasingly complex when the number of targets increases and it is further complicated if the targets are not strong enough to be detected in both spectra.
Many algorithms have been put forward to try to solve this shortcoming, especially in Radar Sensors for ACC systems. The safety and reliability specifications for ACC applications are very strict and therefore the probability of false alarm Pfa, and probability of misdetection(1−Pd), should be kept to very low values. Prior art systems have used algorithms based on the analysis of a large number of hypotheses. The large number of hypotheses result from spectra which are generated from a number of chirps with various slopes (positives, negatives, null). These numerous and complicated algorithms lead to Pfa and (1−Pd) being very dependent on the actual scenarios in the highway, so that in different weather situations and with a large number of objects in the field of view use of the algorithms is questionable if not plainly unsafe.